arXiv subjects

Liu Hao

Publications and source records attributed to Liu Hao.

5 recordsLinked to original sources

arXiv2026

PhoenixNest-Video: Evidence-Grounded Multimodal Agent Framework for Automated Video Interview Assessment

Interview assessment requires per-criterion judgments grounded in behavioral evidence, yet surging applicant volumes have made human-only evaluation costly and inconsistent, while existing AI approaches yield opaque scores without traceable rationale. We introduce PhoenixNest-Video, an evidence-grounded multimodal agent framework for automated video interview assessment. It builds a semantic video graph as structured working memory, performs rubric-conditioned retrieval with cross-modal verification across visual, audio, and textual streams, and produces per-criterion scores anchored to the candidate's materials. A Scorer trained via Rubrics-based Reinforcement Learning with dual rewards for rubric alignment and score-level differentiation internalizes the discriminative structure of multi-level rubrics. PhoenixNest-Video attains 91.50\% grade-level accuracy on VInterview-2025, outperforming substantially larger proprietary models. A compact, rubric-grounded agent therefore scores candidates in closer agreement with an expert panel than direct prompting of much larger models, and exposes the evidence behind each score for human review.

cs.AI↗

arXiv2026

Asteroseismology study of a new faint ZZ Ceti J053009.62+594557.0 discovered in WFST

In this work, we present a detailed asteroseismological analysis of WFST J053009.62+594557.0, a newly discovered faint pulsating white dwarf by the Wide Field Survey Telescope (WFST) with a Gaia G magnitude of 19.13. Analysis of two nights of high-precision WFST g band photometry reveals three significant pulsation frequencies with high signal-to-noise ratios. Follow-up P200/DBSP spectroscopy classifies the object as a DA white dwarf with Teff=11,609 $\pm$ 605 K and M = 0.63$\pm$ 0.22 $M_{\odot}$. To probe its internal structure, we construct asteroseismological models with the White Dwarf Evolution Code (WDEC). After exploring sufficient matching models, best-fitting solutions yield Teff=11,850$\pm$ 10 K and M = 0.600 $\pm$ 0.005 $M_{\odot}$, consistent with independent constraints from Gaia color-magnitude diagram, Gaia XP spectrum, P200 spectral fitting, SED fitting, and Gaia parallax. It has shown that the asteroseismological distance agrees with the Gaia parallax to 1.45\%.

astro-ph.SR↗

arXiv2025

Expanding-and-Shrinking Binary Neural Networks

While binary neural networks (BNNs) offer significant benefits in terms of speed, memory and energy, they encounter substantial accuracy degradation in challenging tasks compared to their real-valued counterparts. Due to the binarization of weights and activations, the possible values of each entry in the feature maps generated by BNNs are strongly constrained. To tackle this limitation, we propose the expanding-and-shrinking operation, which enhances binary feature maps with negligible increase of computation complexity, thereby strengthening the representation capacity. Extensive experiments conducted on multiple benchmarks reveal that our approach generalizes well across diverse applications ranging from image classification, object detection to generative diffusion model, while also achieving remarkable improvement over various leading binarization algorithms based on different architectures including both CNNs and Transformers.

cs.CV↗

arXiv2022

Gradient boosting machines and careful pre-processing work best: ASHRAE Great Energy Predictor III lessons learned

The ASHRAE Great Energy Predictor III (GEPIII) competition was held in late 2019 as one of the largest machine learning competitions ever held focused on building performance. It was hosted on the Kaggle platform and resulted in 39,402 prediction submissions, with the top five teams splitting $25,000 in prize money. This paper outlines lessons learned from participants, mainly from teams who scored in the top 5% of the competition. Various insights were gained from their experience through an online survey, analysis of publicly shared submissions and notebooks, and the documentation of the winning teams. The top-performing solutions mostly used ensembles of Gradient Boosting Machine (GBM) tree-based models, with the LightGBM package being the most popular. The survey participants indicated that the preprocessing and feature extraction phases were the most important aspects of creating the best modeling approach. All the survey respondents used Python as their primary modeling tool, and it was common to use Jupyter-style Notebooks as development environments. These conclusions are essential to help steer the research and practical implementation of building energy meter prediction in the future.

cs.LG↗

arXiv2022

WISE view of changing-look AGNs: evidence for a transitional stage of AGNs

The discovery of changing-look active galactic nuclei (CLAGNs) with the significant change of optical broad emission lines (optical CLAGNs) and/or strong variation of line-of-sight column densities (X-ray CLAGNs) challenges the orientation-based AGN unification model. We explore mid-infrared (mid-IR) properties for a sample of 57 optical CLAGNs and 11 X-ray CLAGNs based on the {\it Wide-field Infrared Survey Explorer} ({\it WISE}) archive data. We find that Eddington-scaled mid-IR luminosities of both optical and X-ray CLAGNs stay just between low-luminosity AGNs (LLAGNs) and luminous QSOs. The average Eddington-scaled mid-IR luminosities for optical and X-ray CLAGNs are $\sim 0.4$\% and $\sim 0.5$\%, respectively, which roughly correspond the bolometric luminosity of transition between a radiatively inefficient accretion flow (RIAF) and Shakura-Sunyaev disk (SSD). We estimate the time lags of the variation in the mid-IR behind that in the optical band for 13 CLAGNs with strong mid-IR variability, where the tight correlation between the time lag and the bolometric luminosity ($τ- L$) for CLAGNs roughly follows that found in the luminous QSOs.

astro-ph.HE↗